The goal of my blog was never to accuse anyone of anything. I try to stay objective and let the reader draw their own conclusion. I want people to understand what’s behind a given tool or the way it’s implemented, so they make decisions consciously, not based on PR. Thanks, Fabrice. I agree with you on a lot of it.
I’d say it’s less of a nightmare and more something to go in with your eyes open. The risks are already there, you just want to understand them before you flip it on.
The lock-in mechanic is the sharper thesis here than the hypocrisy read already in the thread, and there's a number for it. A CSA survey of 1,000 executives this month found 71% report difficulty switching their primary AI vendor, and 91% can't map their own AI dependencies well enough to know how exposed they are. What you're describing with Tag is a specific case of something vendor-risk research is starting to track as its own pattern — an approved vendor quietly routing data to a new AI model through a feature update instead of a fresh risk review. The leverage a buyer actually has isn't at the signature; it's at the moment the agent stops being optional and starts being load-bearing, and for Claude Tag that window already closed.
$dayjob uses WebEx rather than slack but I have no doubt that it works the same. Definitely considering our own chat instance. The convenience of having something summarize meetings is going to be hard to replace though. Yet another reason to have your own personal ai
I spent 12 years at Salesforce on the product side. I know how those decisions are made.
You are analyzing it from a logical standpoint. You need to put a business lens.
Salesforce striked a strategic partnership deal with Anthropic, that’s it.
To put it in perspective it’s similar to Salesforce listing all ISV apps from AppExchange on AWS Marketplace. It’s a strategic partnership.
The PR Saleaforce feeds (privacy) is just PR. I sat in meetings creating a lot of those.
Never take the PR and marketing at face value. Never.
The goal of my blog was never to accuse anyone of anything. I try to stay objective and let the reader draw their own conclusion. I want people to understand what’s behind a given tool or the way it’s implemented, so they make decisions consciously, not based on PR. Thanks, Fabrice. I agree with you on a lot of it.
It was a great post Denis, no issue at all. I was just trying to give a glimpse at how decisions are made
A nightmare.
I’d say it’s less of a nightmare and more something to go in with your eyes open. The risks are already there, you just want to understand them before you flip it on.
The lock-in mechanic is the sharper thesis here than the hypocrisy read already in the thread, and there's a number for it. A CSA survey of 1,000 executives this month found 71% report difficulty switching their primary AI vendor, and 91% can't map their own AI dependencies well enough to know how exposed they are. What you're describing with Tag is a specific case of something vendor-risk research is starting to track as its own pattern — an approved vendor quietly routing data to a new AI model through a feature update instead of a fresh risk review. The leverage a buyer actually has isn't at the signature; it's at the moment the agent stops being optional and starts being load-bearing, and for Claude Tag that window already closed.
$dayjob uses WebEx rather than slack but I have no doubt that it works the same. Definitely considering our own chat instance. The convenience of having something summarize meetings is going to be hard to replace though. Yet another reason to have your own personal ai